Regional Differences in Incident Prefrailty and Frailty

Abstract

Background and Objectives:

The extent to which greater frailty among American compared with European women reflects individual-level characteristics has not been well studied. To test the hypothesis that cardiometabolic conditions and depression and anxiety confound the relationship between region and incident prefrailty and frailty in American compared with European women.

Materials and Methods:

The Global Longitudinal Study of Osteoporosis in Women (GLOW) is a 5-year observational cohort study of women aged ≥55 years. A total of 19,674 participants from the United States and Europe were nonfrail at baseline and provided information on characteristics, including body mass index, depression and anxiety, and cardiovascular disease. We used multivariable Cox proportional hazards models to examine the relationship between region and incident frailty and prefrailty.

Results:

Over 40% of respondents became prefrail or frail during follow-up. Adjusting for age, body mass index, depression and anxiety, cardiovascular disease, and other health-related characteristics, European respondents had a decreased risk of developing prefrailty (2-year hazard ratio [HR]: 0.78, 95% confidence interval [CI]: 0.73–0.84; 3-year HR: 0.74, 95% CI: 0.67–0.81) and frailty (2-year HR: 0.65, 95% CI: 0.56–0.76; 3-year HR: 0.82, 95% CI: 0.68–0.99) compared with American respondents. Risk of incident frailty and prefrailty did not vary by region at 5 years of follow-up.

Conclusions:

Cardiometabolic conditions and depression and anxiety did not account for increased frailty and prefrailty onset among American compared with European women. Differences in smaller regions and environmental characteristics may contribute to frailty and prefrailty.

Introduction

Frailty—a state of increased vulnerability resulting from decreased health reserves, diminished resistance to stressors,¹ and multisystem impairment²—predicts future health decline.^1,3,4 In the Global Longitudinal Study of Osteoporosis in Women (GLOW), American women were more likely to have prevalent frailty or prefrailty compared with women from countries combined across Europe.⁴ The frailty disadvantage in American women compared with European women may contribute to the widening gap in life expectancy at age 50 years between the United States and individual European countries,⁵ as well as to a higher prevalence of functional limitation and disability among American compared with women of individual European countries.⁶

As a potentially modifiable state, frailty is a valuable public health target.⁷ Understanding why European women are less likely to be frail than American women will inform frailty prevention and management in American women and may ultimately prevent or delay outcomes that follow frailty onset. Reasons for differences in frailty between the United States and Europe have not been adequately explored.⁸ Obesity,^3,9 depression,¹⁰ and cardiovascular disease¹¹ are related to frailty and are more common among American than European women.^5,12
–14 However, few studies have directly compared individual-level risk factors and incident frailty in the United States and Europe. Examining onset of frailty, rather than prevalent frailty, more accurately identifies temporal ordering of risk factors for, rather than consequences of, frailty.

The GLOW is a longitudinal survey of women in the United States and Europe that is unique in using identical measures of patient characteristics and frailty across this combination of countries. We examined differences between women from the United States and Europe in the GLOW in incident prefrailty and frailty and the extent to which these differences related to individual health characteristics. We tested the hypothesis that differences in body mass index, depression and anxiety, and cardiovascular disease confound the association between geographic region and incident prefrailty and frailty among women from the United States and Europe.

Methods

The GLOW is an observational study of women aged ≥55 years that examined risk factors for and health consequences of fragility fracture.¹⁵ Briefly, 60,393 women aged ≥55 years from 17 study sites in Australia, Belgium, Canada, France, Germany, Italy, the Netherlands, Spain, the United Kingdom, and the United States have participated in GLOW, with baseline in 2006–2008. For study administrative reasons relating to participant confidentiality, consideration of individual European countries and regions within the United States was not possible. Of these participants, 25,334 were from Europe and 28,170 were from the United States. Regional study sites were recruited from all available physician practices or through primary care networks. Eligible women had visited the practice during the past 2 years. Those who were unable to complete the survey because of cognitive impairment, poor health, or institutionalization were excluded. Participants completed follow-up surveys at 2, 3, and 5 years after baseline. Reason for study dropout was not recorded.

Covariates

In addition to the main confounding variables of interest, body mass index, depression and anxiety, and cardiovascular disease, we included additional characteristics that may confound the association between region and the outcomes of prefrailty and frailty. We considered age as a continuous variable. Respondents provided information on body mass index, depression and anxiety, and cardiovascular disease and related sociodemographic and health variables at baseline. Respondents from France, Italy, Spain, the United Kingdom, and the United States completed comparable questions about education, which we categorized as primary/lower secondary/middle school, higher secondary, and postsecondary. Binary variables included current smoking and the EuroQol-5D (EQ-5D) depression scale¹⁶ (not anxious or depressed vs. moderately or extremely anxious or depressed).

We categorized body mass index as <18.5, 18.5–24.9, 25.0–29.9, and ≥30.0 kg/m² and alcohol consumption as 0, <7, 7–13, 14–19, and ≥20 drinks/week. Respondents reported on history of hypertension, heart disease, osteoarthritis, rheumatoid arthritis, stroke, Parkinson's disease, multiple sclerosis, and cancer; any fracture since 45 years; and current use of antiosteoporosis medication.

Outcomes

Frailty status was assessed at baseline and follow-up years 2, 3, and 5 based on the Fried model components,¹ using an adaptation from the Women's Health Initiative without physical performance tests.^3,4 The sum of points for slowness and weakness (Medical Outcomes Study 36-item Short Form Survey [SF-36] physical functioning scale), poor endurance and exhaustion (from the SF-36 vitality component), and physical activity (self-reported number of days walking ≥20 minutes in past 30 days) ranged from 0 to 5. A score of 0 denoted no frailty. A score of 1 or 2 denoted prefrailty, an intermediate stage between no frailty and frailty, and 3, 4, or 5 indicated frailty.

Statistical analysis

Participants who were nonfrail at baseline and provided information on all baseline exposure variables and who had information about frailty status during at least two consecutive waves among baseline and follow-up years 2, 3, and 5 were included in the analysis sample.

We described baseline characteristics by region, using the maximum available sample size. We described categorical variables as proportions, using the chi-square test or Mantel–Haenszel chi-square test for ordered categories. We described distributions of continuous variables with medians and 25th and 75th percentiles with a Kruskal–Wallis test. Cumulative rates for prefrailty and frailty were calculated within each region using the Kaplan−Meier method; as there was one survey per year there is one time point per year. Separate Cox proportional hazards models were fit for prefrailty and for frailty among nonfrail women at baseline to the outcome of study year after baseline in which prefrailty or frailty first occurred.

In these models, we included only respondents with information on all baseline covariates. Women remaining nonfrail contributed to the model until the end of their follow-up (either a gap in their follow-up or the study completion). Women with an event remained in the analysis only if no follow-up gap occurred before that event. As the data did not meet the proportional hazards assumption, we present the hazard ratio (HR) for each year of follow-up (2, 3, and 5 years) separately. This was done by including an interaction term for region with year in each model.

The first model for each outcome was adjusted for age only. The second model was then fit based on a backward stepwise procedure beginning with clinically relevant variables that were statistically different (p < 0.20) on a univariate level between those who became prefrail at any point during follow-up and the group who remained nonfrail during follow-up. This process was repeated for the outcome of frailty compared with nonfrailty. For both the frailty and the prefrailty models, this starting list comprised depression and anxiety; body mass index; alcohol consumption; history of hypertension, heart disease, osteoarthritis, rheumatoid arthritis, stroke, Parkinson's disease, multiple sclerosis, or fracture; and antiosteoporosis medication use.

For each outcome, variables significant at p < 0.05 were retained in a preliminary multivariable model. Each variable removed during backward selection was then individually added back into the model, and the final model included any variable significant at p < 0.05. If a participant became prefrail and frail during follow-up, we included her in both models. These participants would contribute an event for both the analysis considering prefrailty as an outcome and the analysis considering frailty as an outcome. For the outcome of prefrailty, sensitivity analyses excluded women who became both prefrail and frail during follow-up. Additional sensitivity analysis included education as available as a candidate variable for the backward stepwise procedure. Analyses were conducted in SAS version 9.4 (SAS Institute, Inc., Cary, NC). A grant from Warner Chilcott and Sanofi to the Center for Outcomes Research, University of Massachusetts, supported this study. The University of Maryland, Baltimore, Institutional Review Board approved this study.

Results

A total of 19,674 nonfrail women at baseline (9947 European participants and 9727 American participants) provided information on all baseline covariates and frailty status during at least two consecutive questionnaires between baseline and follow-up. Of the European participants, 81% remained in the analysis sample through year 5, while 11% remained in the sample until year 3, and 8.8% until year 2. Of the American participants, 78% remained in the analysis sample through year 5, and 14% and 7.5% remained in the analysis sample until years 3 and 2, respectively. Participants remaining in the analysis sample only until follow-up years 2 or 3 were more likely to be moderately or extremely anxious or depressed, with a body mass index of ≥30 kg/m², and to have a history of hypertension or heart disease than participants who completed surveys in year 5 (results not presented).

The cumulative prefrailty incidence rate was greater in the United States than Europe at follow-up years 2 (24% vs. 19%), 3 (39% vs. 30%), and 5 (50% vs. 42%) (Table 1). The cumulative frailty incidence rate was also greater in the United States than Europe at all follow-up time points (8.7% for the United States vs. 5.2% for Europe in year 2, 14% in the United States vs. 9.3% in Europe in year 3, and 21% for the United States vs. 15% for Europe in year 5).

Table 1.

Cumulative Prefrailty and Frailty Incidence Rates by Follow-Up Year and Region

	Europe (%)	Europe, n	United States (%)	United States, n	p^a
Prefrailty					<0.0001
Year 2	1787 (19)	9551	2223 (24)	9257
Year 3	1014 (30)	7114	1267 (39)	6565	<0.0001
Year 5	936 (42)	5450	816 (50)	4605
Frailty
Year 2	347 (5.2)	6703	532 (8.7)	6098
Year 3	251 (9.3)	5706	314 (14)	5097
Year 5	291 (15)	4805	301 (21)	4090

Number of incident cases, cumulative incidence rate, and number of respondents at risk presented per year of follow-up.

Log rank test for equality over strata.

Participants from the United States were more likely to have completed higher education, to have a body mass index ≥30 kg/m², a history of hypertension or cancer, and be current users of antiosteoporosis medications compared with European participants (Table 2). European participants were more likely to be current smokers, to have moderate or extreme depression or anxiety, and to have osteoarthritis than American participants.

Table 2.

Baseline Sample Characteristics by Region (n = 19,674)

	Europe (n = 9947)^a	United States (n = 9727)^a	p
Age (years), median (IQR)	65 (60–71)	66 (61–72)	<0.0001
Education,^b n (%)			<0.0001
Primary/lower secondary/middle school	2694 (49)	49 (0.5)
Higher secondary	1481 (27)	1995 (21)
Postsecondary	1318 (24)	7624 (79)
Current smoking, n (%)	977 (9.9)	442 (4.6)	<0.0001
EQ-5D depression scale, n (%)			<0.0001
Not anxious or depressed	6021 (61)	7464 (77)
Moderately or extremely anxious or depressed	3839 (39)	2196 (23)
Body mass index, n (%)			<0.0001
<18.5 kg/m²	158 (1.6)	162 (1.7)
18.5–24.9 kg/m²	5025 (52)	4595 (48)
25.0–29.9 kg/m²	3302 (34)	3168 (33)
≥30 kg/m²	1126 (12)	1657 (17)
Alcohol consumption, n (%)			<0.0001
0 drinks/week	3509 (35)	4041 (42)
<7 drinks/week	4338 (44)	3849 (40)
7–13 drinks/week	1637 (16)	1466 (15)
14–19 drinks/week	384 (3.9)	323 (3.3)
≥20 drinks/week	58 (0.6)	32 (0.3)
Medical history, n (%)
Hypertension	3558 (36)	4436 (46)	<0.0001
Heart disease	829 (8.5)	779 (8.1)	0.40
Osteoarthritis	3992 (41)	2340 (24)	<0.0001
Rheumatoid arthritis	24 (0.3)	48 (0.5)	0.005
Stroke	164 (1.7)	178 (1.8)	0.34
Parkinson's disease	21 (0.2)	20 (0.2)	0.92
Multiple sclerosis	21 (0.2)	33 (0.3)	0.09
Cancer	921 (9.3)	1553 (16)	<0.0001
Any fracture since age 45 years, n (%)	1788 (18)	1800 (19)	0.40
Antiosteoporosis medication use, n (%)	1130 (12)	2308 (24)	<0.0001

Denotes maximum sample size and sample size for age; largest available sample per variable used.

France, Italy, Spain, United Kingdom, and United States.

EQ-5D, EuroQol-5D; IQR, interquartile range.

In age-adjusted models, European participants were less likely to become prefrail (HR: 0.80, 95% confidence interval [CI]: 0.76–0.86, Table 3) or frail (HR: 0.67, 95% CI: 0.58–0.76) by year 2 than American participants. Similar age-adjusted relationships remained at year 3 (prefrailty HR: 0.76, 95% CI: 0.70–0.83; frailty HR: 0.80, 95% CI: 0.68–0.95). Adjusting for body mass index, depression and anxiety, hypertension, heart disease, and additional covariates identified as potential confounding variables (alcohol consumption, osteoarthritis, rheumatoid arthritis, Parkinson's disease, and fracture since age 45 years for the prefrailty and frailty outcomes plus history of cancer and stroke for the frailty outcome) in these time intervals did not substantially alter relationships. At 5 years of follow-up, region was not related to risk of prefrailty or frailty onset. Sensitivity analyses that included education and, separately, that excluded from models with the outcome of prefrailty participants who became prefrail and then frail produced similar results.

Table 3.

Associations Between Region and Prefrailty and Frailty Onset

	Age-adjusted HR for region^a (95% CI)	p^b	Multivariable HR^c for region^a (95% CI)	p^b
Prefrailty	N = 18,808		N = 16,273
Year 2	0.80 (0.76–0.86)	<0.0001	0.78 (0.73–0.84)	<0.0001
Year 3	0.76 (0.70–0.83)	<0.0001	0.74 (0.67–0.81)	<0.0001
Year 5	1.00 (0.91–1.10)	0.99	0.94 (0.85–1.05)	0.26
Frailty	N = 12,801		N = 11,025
Year 2	0.67 (0.58–0.76)	<0.0001	0.65 (0.56–0.76)	<0.0001
Year 3	0.80 (0.68–0.95)	0.009	0.82 (0.68–0.98)	0.032
Year 5	0.93 (0.79–1.09)	0.34	0.90 (0.76–1.07)	0.24

Europe versus United States.

From a Wald chi-squared test.

Multivariable models adjusted for age, body mass index category, depression and anxiety, hypertension, heart disease, alcohol consumption, osteoarthritis, rheumatoid arthritis, Parkinson's disease, and fracture since age 45 years. The frailty model additionally adjusted for cancer and stroke.

CI, confidence interval; HR, hazard ratio.

Discussion

In a prospective study of women aged ≥55 years, European women were less likely to become prefrail or frail than American women at 2 and 3 years of follow-up. Differences persisted after adjustment for baseline characteristics of body mass index, depression and anxiety, hypertension, and heart disease, in addition to other potential confounding variables. Differences in prefrailty and frailty onset by region were not present at 5 years of follow-up. Results did not support our hypothesis that advantages among European women in body mass index, depression and anxiety, and cardiovascular disease compared with American women confound the association between geographic region and greater incident prefrailty and frailty among women from the United States and Europe. In addition, European women did not have advantages in heart disease or depression and anxiety compared with American women. Previous cross-sectional studies could not address the importance of geographic region for frailty and prefrailty onset.⁸ The relationship between region and risk for incident frailty or prefrailty weakened over time, suggesting that location in the United States versus Europe is only influential among those most immediately susceptible to frailty or prefrailty. Analysis sample attrition over time was similar for American and European women and was unlikely to contribute strongly to this diminishing trend over time. Our direct comparison of frailty and individual characteristics of women from the United States and Europe using the same measures adds to the growing understanding of health differentials between these two populations.

Previous studies of geographic differences among European countries using a similar frailty measure based on the Fried model have been cross sectional⁸ or with short follow-up¹⁷ and could not differentiate temporal ordering of frailty and related patient characteristics, which has implications for the timing and target of intervention. These studies along with a longitudinal study of worsening frailty based on the Fried model¹⁸ found differences in frailty across European countries, although comparisons were unadjusted or adjusted only for individual sociodemographic characteristics. We extended such findings by adjusting for health-related individual-level characteristics and by including comparison to a country outside of Europe.

Unmeasured individual characteristics or social, built, or physical environmental factors¹⁹ may influence the difference in frailty and prefrailty incidence between American and European women. GLOW was not able to capture some individual characteristics related to frailty that are more prevalent in the United States than Europe, such as diabetes.^6,19,20 Social aspects of the environment are known frailty determinants,^21,22 and lack of such details may limit the ability to explain differences in frailty and prefrailty incidence between the United States and Europe.²³ In a previous European study, level of prevalent frailty across countries was related to socioeconomic environmental factors and social policy.^24,25 In this study, European women continued to have lower prefrailty and frailty incidence after accounting for education. Education may not fully capture social environment in the present or from earlier in the life course, both of which may have implications for later life frailty.⁶

We note several limitations to our study. For administrative reasons, we were unable to consider individual European countries, which are heterogeneous in terms of frailty^8,24 and individual health characteristics of obesity,¹² depression,^5,12
–14 and heart disease,^5,12
–14 as well as socioeconomic environmental factors and social policy.^24,25 Despite this limitation of grouping European countries together, this analysis provides useful insight as one of the first to directly include countries both within and outside of Europe using identical study questions for incident, rather than prevalent, prefrailty, frailty, and a diverse set of social and health-related risk factors. Other methodological considerations may also limit the analysis. To include a variety of women aged ≥55 years, women with any activity in the past 2 years in primary care practices were invited to participate in GLOW.¹⁵ However, regional differences in healthcare access may contribute to recruitment bias. In addition, physicians who elected to participate in the survey may not represent all physicians in the area, and study participants may have more interest in their health than those who declined to participate.¹⁵ Our survey did not capture reason for study attrition. However, participants remaining in the study sample only until follow-up years 2 or 3 were more likely to have baseline health characteristics related to incident frailty and prefrailty, compared with participants who remained in the study sample until follow-up year 5. This baseline profile also means these respondents were more likely to drop out due to death or not complete the survey because of cognitive impairment, poor health, or institutionalization. If these other events occurred before the onset of frailty or prefrailty, our analysis may underestimate the incidence of outcomes.

By censoring eligible participants who dropped out of the survey in our models, we still captured risk of frailty for as long as possible among these participants. As only respondents from the United States reported on race, we could not consider whether race may also confound the relationship between geographic region and incident prefrailty and frailty. In analysis of the Women's Health Initiative adjusting for individual-level social and health characteristics, black women were less likely to develop frailty but equally likely to develop prefrailty compared with white women, in 3 years of follow-up.³ Race may capture additional differences in individual or environmental characteristics that relate to incident prefrailty or frailty risk. To our knowledge, studies of frailty in individual European counties have not included race.^8,15,18

Greater incident prefrailty and frailty in American compared with European women were not related to the European advantage in obesity or hypertension. This study adds to the growing understanding of how differences in health and mortality develop and persist among women in the United States versus Europe^5,6 by demonstrating that differences in individual-level health characteristics may not be the most effective preventive targets. Future studies should directly address environmental factors and individual European countries that contribute to greater onset of prefrailty and frailty in American compared with European women.

Footnotes

Acknowledgments

We thank the physicians and project coordinators who participated in GLOW. Sophie Rushton-Smith, PhD, provided editorial and project management support. The study was supported by a grant from Warner Chilcott and Sanofi to the Center for Outcomes Research, University of Massachusetts (Worcester, MA).

Author Disclosure Statement

Dr. Tom received grant support from Novartis. Dr. Adachi received consulting fees or other remuneration from Amgen, Eli Lilly, Merck, Novartis, Warner Chilcott; research grants from Amgen, Eli Lilly, Merck, and Novartis; nonremunerative position of influence on the IOF Board of Directors, Osteoporosis Canada; speakers bureaus for Amgen, Eli Lilly, Merck, Novartis, and Warner Chilcott. Dr. Anderson received funding from The Alliance for Better Bone Health (Sanofi-Aventis and Warner Chilcott) and Pfizer. Dr. Chapurlat received funding from the French Ministry of Health, Merck; honoraria from Amgen, Servier, Novartis, Lilly, Roche, Pfizer, BMS, Bioiberica; and is an Advisory Board member for Amgen, UCB, Bioiberica. Dr. Compston received lecture fees from Servier and Amgen and grant support from Amgen, The Alliance for Better Bone Health (Sanofi-Aventis and Warner Chilcott), and Eli Lilly. Dr. Cooper previously consulted for/received lecture fees from Amgen, The Alliance for Better Bone Health (Sanofi-Aventis and Warner Chilcott), Lilly, Merck, Servier, Novartis, and Roche-GSK. Dr. Díez-Pérez received consulting fees and lectured for Eli Lilly, Amgen, GSK, and Merck; consults for/is an Advisory Board member for Eli Lilly and Amgen; shareholder Active Life Scientific. Dr. Gehlbach received funding from Pfizer. Dr. Greenspan previously consulted/been an Advisory Board member for Amgen, Lilly, and Merck; and received grant support from The Alliance for Better Bone Health (Sanofi-Aventis and Proctor & Gamble), Amgen, and Lilly. Dr. Hooven received funding from Pfizer. Dr. March has been an Advisory Board member for Servier; received speakers' bureau fees and support to travel to scientific meetings from Servier, Merk, Amgen, UCB, and Pfizer. Dr. Netelenbos previously consulted for Roche Diagnostics, Daiichi-Sankyo, Proctor & Gamble, and Nycomed; received lecture fees, travel and accommodation from E. Lilly, Amgen, Novartis, and Will Farma; grant support from The Alliance for Better Bone Health and Amgen. Dr. Pfeilschifter received funding from The Alliance for Better Bone Health (Sanofi-Aventis and Warner Chilcott), received grant support from MSD, and lecture fees from GlaxoSmithKline. Dr. Roux received honoraria from and consults/is an advisory board member for Alliance, Amgen, Lilly, Merck, Novartis, Nycomed, Roche, GlaxoSmithKline, Servier, and Wyeth. Dr. Saag consulted for or received other remuneration from Merck, Amgen, and Eli Lilly; research grants from Merck; nonremunerative positions of influence on the NOF Board of Trustees and as ACR Chair on the Quality of Care Committee. Dr. Silverman received grant support from Pfizer, Lilly, Amgen, and Medtronic; served on Speakers' Bureaus for Lilly, Amgen, and Pfizer; Advisory Board member for Lilly, Amgen, and Lilly. Dr. Siris previously consulted for Amgen, Lilly, Novartis, Merck, and Pfizer; served on Speakers' Bureaus for Amgen and Lilly. Dr. Watts received honoraria for lectures during the past year from Amgen, Lilly, Novartis, and Warner Chilcott; consulting fees during the past year from Abbott, Amgen, Bristol-Myers Squibb, Endo, Imagepace, Johnson & Johnson, Lilly, Medpace, Merck, Nitto Denko, Noven, Novo Nordisk, Pfizer/Wyeth, and Quark; research support (through his Health System) from Merck and NPS; and cofounded, stock options in and a director of OsteoDynamics. Dr. LaCroix received funding from The Alliance for Better Bone Health (Sanofi-Aventis and Warner Chilcott) and is an Advisory Board member for Amgen. All other authors have no competing financial interests.

References

Fried

, Tangen

, Walston

, et al. Frailty in older adults: Evidence for a phenotype. J Gerontol A Biol Sci Med Sci, 2001; 56:M146–M156.

Rockwood

. Frailty and its definition: A worthy challenge. J Am Geriatr Soc, 2005; 53:1069–1070.

Woods

, LaCroix

, Gray

, et al. Frailty: Emergence and consequences in women aged 65 and older in the Women's Health Initiative Observational Study. J Am Geriatr Soc, 2005; 53:1321–1330.

Tom

, Adachi

, Anderson

Jr , et al. Frailty and fracture, disability, and falls: A multiple country study from the Global Longitudinal Study of Osteoporosis in Women (GLOW). J Am Geriatr Soc, 2013; 61:327–334.

Glei

, Meslé

, Vallin

. Diverging trends in life expectancy at age 50: A look at causes of death. In: Crimmins

, Preston

, Cohen

, eds. International differences in mortality at older ages: Dimensions and sources. Washington, DC: The National Academies Press, 2010:17–67.

Crimmins

, Garcia

, Kim

. Are international differences in health similar to international differences in life expectancy? In: Crimmins

, Preston

, Cohen

, eds. International differences in mortality at older ages: Dimensions and sources. Washington, DC: The National Academies Press, 2010:68–104.

Cameron

, Fairhall

, Langron

, et al. A multifactorial interdisciplinary intervention reduces frailty in older people: Randomized trial. BMC Med, 2013; 11:65.

Santos-Eggimann

, Cuenoud

, Spagnoli

, Junod

. Prevalence of frailty in middle-aged and older community-dwelling Europeans living in 10 countries. J Gerontol A Biol Sci Med Sci, 2009; 64:675–681.

Hubbard

, Lang

, Llewellyn

, Rockwood

. Frailty, body mass index, and abdominal obesity in older people. J Gerontol A Biol Sci Med Sci, 2010; 65:377–381.

10.

Lakey

, LaCroix

, Gray

, et al. Antidepressant use, depressive symptoms, and incident frailty in women aged 65 and older from the Women's Health Initiative Observational Study. J Am Geriatr Soc, 2012; 60:854–861.

11.

Garcia-Garcia

, Gutierrez Avila

, Alfaro-Acha

, et al. The prevalence of frailty syndrome in an older population from Spain. The Toledo Study for Healthy Aging. J Nutr Health Aging, 2011; 15:852–856.

12.

World Health Organization.. WHO global infobase. Available at: https://apps.who.int/infobase/Comparisons.aspx Accessed January 6, 2015 .

13.

Kromhout

. Epidemiology of cardiovascular diseases in Europe. Public Health Nutr, 2001; 4:441–457.

14.

WHO International Consortium in Psychiatric Epidemiology. Cross-national comparisons of the prevalences and correlates of mental disorders. Bull World Health Organ, 2000; 78:413–426.

15.

Hooven

, Adachi

, Adami

, et al. The Global Longitudinal Study of Osteoporosis in Women (GLOW): Rationale and study design. Osteoporos Int, 2009; 20:1107–1116.

16.

Brazier

, Walters

, Nicholl

, Kohler

. Using the SF-36 and EuroQol on an elderly population. Qual Life Res, 1996; 5:195–204.

17.

Castell

, van der Pas

, Otero

, Siviero

, et al. Osteoarthritis and frailty in elderly individuals across six European countries: Results from the European Project on OSteoArthritis (EPOSA). BMC Muscuoloskelet Disord, 2015; 16:359.

18.

Etman

, Burdorf

, Van der Cammen

TJM

, Mackenbach

, VanLenthe

. Socio-demographic determinants of worsening in frailty among community-dwelling older people in 11 European countries. J Epidemiol Community Health, 2012; 66:1116–1121.

19.

Kalyani

, Tian

, Xue

, et al. Hyperglycemia and incidence of frailty and lower extremity mobility limitations in older women. J Am Geriatr Soc. 2011; 60:1701–1707.

20.

Blaum

, Xue

, Tian

, et al. Is hyperglycemia associated with frailty status in older women?. J Am Geriatr Soc, 2009; 57:840–847.

21.

Macintyre

, Ellaway

, Cummins

. Place effects on health: How can we conceptualise, operationalise and measure them?. Soc Sci Med, 2002; 55:125–139.

22.

Lang

, Hubbard

, Andrew

, Llewellyn

, Melzer

, Rockwood

. Neighborhood deprivation, individual socioeconomic status, and frailty in older adults. J Am Geriatr Soc, 2009; 57:1776–1780.

23.

King-Kallimanis

, Kenny

, Savva

. Factor structure for the frailty syndrome was consistent across Europe. J Clin Epidemiol, 2014; 67:1008–1015.

24.

Theou

, Brothers

, Rockwood

, Haardt

, Mitnitski

, Rockwood

. Exploring the relationship between national economic indicators and relative fitness and frailty in middle-aged and older Europeans. Age Ageing, 2013; 42:614–619.

25.

Sirven

. On the socio-economic determinants of frailty: Findings from panel and retrospective data from SHARE. IRDES working paper. Available at: www.irdes.fr/EspaceAnglais/Publications/WorkingPapers/DT52SocioEconomicDeterminantsFrailty.pdf Accessed January 6, 2015 .